POLISH JOURNAL OF ECOLOGY 48 Suppi. 87-96 2000 (Pol. J. Ecol.)

. Bank vole biology: Recent advances In• the population biology of a model spectes

Heikki HENTTONEN

Vantaa Research Center, Finnish Forest Research Institute, POB 18, FIN-010301 Vantaa, Finland, e-m ail: [email protected]

LONG-TERM DYNAMICS OF THE BANK VOLE •• CLETHRIONOMYS GLAREOLUS AT PALLASJARVI, NORTHERN FINNISH TAIGA

ABSTRACT: The whole rodent community feature in northern Fennoscandia. These cy­ (eight arvicolinc species) has been followed at cles are, in turn, reflected in many other com­ Pallasjarvi. at the northern limit of Clethrionon1ys ponents of northern nature, like small game glareolus in Finnish Lapland, since 1970. Dynamics and vegetation (Hansson and Henttonen \vere cyclic until the mid 1980 's~ but since then the 1989). The very special feature of the Pallas pattern has been stable. Also the species abundan­ rodent assemblage is the high number of syn­ ces have changed. The delayed density dependence, characterising the cyclic period, is not found during topic species (Hent tonen et al. 1977). The the stable period. Causes for this change in cyclicity rodent fauna includes three species of Cle­ are discussed. The bank vole is the most common thrionomys, bank vole C. glareolus, red vole rodent species in forests up to its northern limit. The C. rutilus and grey-sided vole C. rufocanus, long-term, year around live trapping studies and fe­ two species of Microtus, field vole M agres­ eding experiments suggest e.g. that delayed matura­ tis and root vole M oeconomus, two species tion of young is not optimal but due to social con­ of lemmings, Norwegian lemming Lemmus straints. Food addition resulted in higher densities. lemmus and wood lemming Myopus schisti­ but the effect on the density-dependent structure color, and the water vole Arvicola terrestris. \vas negligible and the dynamics \Vere not affected by food addition. Also the muskrat Ondatra zibethicus is occa­ KEY-WORDS: bank vole, Clethrionomys gla­ sionally observed. These arvicoline species reolus, arvicoline species, cycles, Lapland, densi­ have, however, different habitat characteris­ ty-dependence, interspecific competition tics and diets. In addition, the small mammal fauna consists of four species of Sorex shrews, S. araneus, S. caecutiens, S. minutus 1. INTRODUCTION and S. minutissimus, and the water shrew Neomys fodiens. The Pallasjarvi area (68°30'N, 24°09'E), The high number of species allows for in the southern part of the Pallas-Ounas Na­ studying dynamical processes at various lev­ tional Park in western Finnish Lapland, has els, i.e. the whole small mammal community, been a core place for an intensive rodent re­ the rodent community and population level. search since 1970. There are several reasons As a consequence, it is possible to differenti­ why this area in particular is so suitable for ate between the factors affecting simultane­ long-term ecological research on voles and ously the whole small mammal or rodent as­ lemmings. The population cycles of voles semblage or only the population of single and lemmings have been a very characteristic species (Henttonen 1985, Henttonen 88 H. Henttonen

1987, Hansson and Henttonen 1988). Ba­ 15 x 15 m with three snap traps at each corner sically this is the difference between extrinsic (Myllymaki et al. 1971)) on each of the vs. intrinsic population regulation (Hent­ main habitats have been used. In old taiga tonen et al. 1987), i.e. predation, food and forests and peatlands, trapping is done in diseases against social stress and behavioural groups of 3-7 quadrats. In each group, the mechanisms (self regulation), even though distance between quadrats is at least 50 m, species specific resource interaction could af­ usually more. The distance between groups is fect only one species. 0.5-5 km. One large clear cut area is used. The high number of rodent species is Trappings have been done immediately after partly a result ofthe encounter ofnorthern and the snow melt at the turn of May and southern faunal elements, but it may also be June-early June, and again in September. For pattly facilitated by the pronounced popula­ more details, see Henttonen et al. (1987). tion cycles ofthe rodent species (Henttonen The main purpose of these basic moni­ and Hansson 1984, Hanski and Hent­ toring trappings has been to describe the gen­ tonen 1996). The habitat selection of these eral features of the population dynamic pat­ rodent species overlap, and interspecific com­ terns on a larger area, to obtain estimates of petition at least for space during the population population densities on the main habitats, to peaks is a common phenomenon (H e n t - collect data on demographic and reproduc­ t o n e n et al. 1977, H e n t t o n e n and tive parameters, as well as material for parasi­ Hansson 1984, Jortikka 1990, Han­ tological studies (e.g. Henttonen 1985, ski and Henttonen 1996). The deep 1987, 1997, Henttonen et al. 1977, 1987, population crashes affecting all species simul­ Hansson andHenttonen 1985b,Hauki­ taneously could be one factor promoting spe­ salmi et al. 1987, 1988, 1996, Hanski et cies coexistence. In respect of the bank vole, al. 1994). The data from small quadrats on a the Pallasjarvi area has one more specific fea­ larger area serve as background and compari­ ture: the study area is only some 40-50 km son for more detailed live trapping studies on from the northern border ofthe species. more restricted sites (Henttonen et al. 1987, Hanski et al. 1994). 2. EXTENSIVE MONITORINGS AT •• 3. INTENSIVE LIVE-TRAPPING PALLASJARVI •• STUDIES AT PALLASJARVI The Pall asj arvi area is characterised by an isolated mountain group. The forests on Intensive live-trapping studies have been the slopes are characterised by Norway done to follow the populations of individu­ spruce Picea abies, downy birch Betula pu­ ally marked animals. Depending on the pur­ bescens and Scots pine Pinus silvestris. The pose of the specific study, a variable number valleys are covered by pine bogs and mires. ofstudy grids varying in size from 1 to 4.5 ha The dominant feature in the forests is the con­ have been in use. Depending on the number tinuous and dense cover of the bilberry Vac­ of replicates, a 10 x 10m or 15 x 15m grid cinium myrtillus in the field layer and the pattern has been used. Thus, the number of mosses Pleurozium schreberi, Hylocomium live traps per site has usually been 80- 130. splendens and Dicranum spp in the bottom The trapping routine and the number of trap layer. The snow usually melts in late May, checkings per a trapping period has also de­ and pern1anent snow covers prevails since pended on the number of replicated sites. Of late October. The maximum snow cover on course, within a specific experiment or pro­ the study sites is around 1 meter in gram the protocol has been fixed. March- April. The following main live trapping proj­ The basic monitoring on three main ects have been running in old taiga forests. habitats (old taiga forests, bogs and mires, 1) The large control grid (4.5 ha) was and clear cuts) is done with snap trapping on started in 1977, and since spring 1980 it has the permanent sites twice a year. Additional been yearly trapped several times from spring trappings have been done at the tree line and to autumn. During 1982- 1992 the grid was in­ in the alpine zone. The results on the bank tensively live trapped year around. 10 x 10 m voles in this atticle are presented primarily grid pattern was used but at each moment traps from taiga forests. Characterisation of other were situated at 20-m intervals. In snow free habitats can be found in Henttonen et al. seasons, a trapping period consisted of 16 trap ( 1987). 30 permanent small quadrats (grids of checkings at 6-hour intervals. Each point was Long-term dynamics of the bank vole 89 trapped for 24 hours and then the trap was voles were medicated with an antihelminthic moved 10 m to the next point. During snowy (see Haukisalmi and Henttonen 2000). seasons (late October to late May) permanent 4) Two meadows dominated by the root trap chimneys were used at 20-m intervals. In voles have been live trapped from 1981 to winter, trapping was done in day time starting 1996 and have been snap-trapped twice a at 8 a.m. and traps were checked at 2-3 hour year since then. intervals until 5-6 p.m. It should be noticed The live-trapping studies have had (at that in winter time bank voles are day active least) the following purposes: (though twilight in mid winter lasts only for -To give real density values in compari­ few moments). Due to this very intensive trap­ son to estimates obtained with snap­ ping schedule, the sampling error is very low, trappings. Live trappings are much more la­ only about 2% (Yoccoz et al,. in press). bourous to do than snap-trappings, and there­ 2) The large food grid (2.8 ha) was fore intensive live-trapping data can be ob­ started in spring 1982 and live-trapping run tained only in restricted sites. As a contrast, identically to the large control grid until snap-trappings can cover larger areas, but autumn 1996. The first experiment with addi­ snap-trappings give only indices. Combining tional food (oat seeds) lasted from June 1982 these two approaches is fruitful (Henttonen to September 1996. The second experiment et al. 1987, Hanski et al. 1994). with oat seeds and mouse chow ran from - In experiments with supplemental spring 1998 to autumn 1992. Details ofthese food, the role of food in the population dy­ two grids and projects have been given by namics of bank voles has been studied (Hent tonen et al. 1987, P rev o t-J u 11 i a r d Henttonen et al. (1987) and Prevot­ et al. 1999, Y o cc o z et al., in press). Julliard et al. (1999). - As told, interspecific competition 3) Replicated live-trapping grids. Nine among the rodent species is common. This is grids, each 1.8 ha with 10 x 10 trapping points especially so between the breeding individuals at 15-m intervals, were established in spring of Clethrionomys species. Breeding females 1986. Traps were permanently at the same ofClethrionomys species have exclusive terri­ points. In winter trap chimneys were used at tories also between the species, and this is di­ every point. In summer time, each trapping pe­ rectly related to the space available to a spe­ riod included 8 trap checkings at 8-hour inter­ cies in accordance with the competitive inter­ vals. In winter, three trap checkings during specific hierarchy (Henttonen et al. 1977, each of the three trapping days were done. To Henttonen and Hansson 1984, J ortikka have replicates, we had to compromise in trap­ 1990, Hanski and Henttonen 1996). ping efficiency and grid size. - We have studied experimentally e.g. In snow free seasons in 1986 and 1987, whether helminth parasites can affect the sur­ the interspecific competition between the vival and dynamics ofthe bank vole by medi­ bank vole and red vole was studied. Once a cating voles in the experimental populations month, bank voles were removed on three in the field. In these experiments individual grids, red voles from three grids, and three perforn1ance and worrn burden (measured by grids remained as controls (Jo rti kka 1990). counting the worrn eggs in the feces of indi­ Starting in spring 1988, a replicated experi­ vidual voles) were monitored. ment with supplemental food (mouse chow) - In addition to rodent and parasite was started on six of these grids (three food monitoring, we have followed the abundance and three control grids). Three remaining C. of small mustelids (least weasels Mustela rutilus grids where bank vole had been re­ nivalis and stoats M erminea). In summer moved in 1987, were monitored in snow free time we have obtained data from live trap­ seasons until 1992 to document the return of pings and winter also from snow trackings interspecific relations to the normal situation. (Henttonen et al. 1987, Oksanen and The replicated food experiment lasted from Henttonen 1996). spring 1988 to autumn 1992, and experiment ran year around. After the experiment with extra food was finished, the food grids were 4. SOME BASIC CONCLUSIONS monitored in snow free season until autumn ON MICROTINE CYCLES.. 1994. In snow free seasons 1993 and 1994 the AT PALLASJARVI three controls ofthe previous experiment and the above mentioned three former C. rutilus The very characteristic feature of the ro­ grids were used in an experiment where bank dent cycles in northern F ennoscandia is (or at 90 H. Henttonen

• C. gla . 35 0 C. rut. ... C. rut. 30 • V' Others 25 ~ +"" ·-(/) 20 • c: Q) • 0 15 • • • • • 10 •

5

0 1970 1975 1980 1985 1990 1995 2000 Years

Fig. 1. Long-term dynamics of arvicoline rodents in old taiga forests at Pallasjarvi, western Finnish Lapland. Others refer primarily to Lemmus lemmus. Data based on snap-trappings twice a year. Density index is number of voles per 100 trap nights least has been) the synchronous deep crash in Even ifthe rodent community as a whole cy­ all species (Fig 1). This is the reason to em­ cled together until mid 1980's, the abun­ phasize the community approach in the re­ dances of individual species changed from search of rodent cycles (H an ss on and cycle to another. Yet, very different densities Henttonen 1985a, 1988, Henttonen of a species during successive cyclic peaks 1987, Henttonen et al. 1987, Henttonen normally lead to similar total crashes. Fur­ and Hanski 2000, Henttonen and Wall­ thermore, the second high density year during gren 2000, Hanski and Henttonen 1996, the peak phase (like in 1970, 1974, 1978, Hanski eta!. 1991, 1993, 1994). These syn­ 1982) does not automatically lead to crash as chronous crashes cannot be explained by in­ found during the extended peak in 1983. In trinsic factors alone. Ofcourse the population other words, direct density-dependent intrin­ increase of rodents can be first slowed down sic mechanisms do not alone cause the cycle by social factors, but this does not mean that (Henttonen et al. 1987). This is further these same factors cause the crash. Why supported by the data from the last 15 years would all species commit a suicide together? when the pattern of rodent dynamics at Pal­ The only reasonable explanation for the syn­ lasjarvi (and in Lapland generally) has chronous crashes is a common extrinsic fac­ changed. Even though the bank voles now of­ tor (Henttonen 1985, Henttonen et al. ten reach the same autumn densities as they 1987, Hansson and Henttonen 1988) and did earlier only during the cyclic peak phases, the most probable factor is the predation by their dynamics remain multi-annually rather small mustelids. In cyclic times at Pallasjarvi, stable. Recent time-series analyses also indi­ this interspecfic synchrony extended also to cate that delayed density-dependence is insectivorous shrews (Hent tone n 1985), prevalent in the long-term data sets on small but has disappeared since then. The role of rodent fluctuations in northern Fennoscandia specialist predation is further corroborated by suggesting that cycles are caused by a trophic studies elsewhere in Fennoscandia (e.g. interaction (Hanski et al. 1993, Stenseth N orrdahl and Korpimaki 1995, Kor­ 1999, Hansen et al. 1999). For the reason pimaki and N orrdahl 1998; it should be re­ told above, the predation by small mustelids membered, however, that in the three-year seems as the most probable cause. However, cycles in western Finland the abundance and for the recent stable period with low weasel role ofavian predators is clealy higher than in abundanace at Pallasjarvi, the analysis of the north, see also H an ski et al. 1991). Y o cco z et al. (in press) showed no detectable Some other features in the Pallasjarvi delayed density-dependence in the bank vole data set speak for community approach, too. dynamics on the large control and food grids. Long-term dynamics of the bank vole 91

In mid 1980's and at the turn of 1980's keeps optimal prey low, below the threshold and 1990's, a drastic change in the dynamics density needed for breeding of least weasels, of small rodents has occurred in northern and least weasels cannot increase. In proper F ennoscandia. Starting in early 1980's the cyclic times the least weasel dominated over small mammal dynamics at Pallasjarvi began stoats in abundance, especially in the late to change and since mid 1980's the dynamics peak-decline phase (Viitala 1977, Hent­ have been very different from the earlier tonen et al. 1987). Now, during more stable ones. After an extended cycle in 1982-84 and rodent dynamics, stoats seem to dominate a the subsequent low phase in 1985, the dy­ (Henttonen eta/. 1987, unpubl., Oksanen namics of the bank vole and red voles in for­ et al. 1999). Our food addition experiments ests have turned to seasonal. The prevalent (Henttonen et al. 1987; Prevot-Julliard trend is the density increase in snow free sea­ et al. 1999, Henttonen, unpubl.) show that in son and decline in winter. The multi-annual the present more stable situation, the impact pattern has disappeared. The rodent cycle has ofsupplemental food on the densities ofvari­ prevailed in northern Fennoscandia at least ous rodent species depends on the suscepti­ for decades. Therefore the change in cyclicity bility ofthe vole species to predation. In other has been quite puzzling. words, supplemental feeding does not in­ The large scale geographic synchrony crease the densities of larger and clumsier broke down, and the dynamics ofsmall agile species, which supposedly are heavily preyed species turned much more stable than earlier by stoats in summer time, but it increases the (e.g. C. glareolus and C. rutilus) while some densities ofsmall agile species like C. glareo­ earlier abundant large and competitively lus and C. rutilus. dominant species have become much less A core question is why the change has common, even rare (e.g. C. rufocanus and M occurred. The model ofHanski and Hent­ agrestis). Similarly, the least weasel M t onen ( 1996) shows that this kind of change nivalis has declines and the stoat M erminea in a system characterized by two competing is more common, or at least dominating now. prey species, which differ in their susceptibil­ Since then this "Pallasjarvi syndrome" has ity to predation, and a shared predator, is fully occurred elsewhere in Lapland (Henttonen possible due to inherent (chaotic?) properties and Wallgren 2000) and in northern Swe­ of the system. The model predictions resem­ den (Hansson 1999). Henttonen (1987) ble very closely what has really happened e.g. and Henttonen et al. (1987) suggested that at Pallasjarvi and Kilpisjarvi. A dynamic the simultaneous long-term phase oflow den­ change from one state to another may happen sities in M agrestis, the favoured foor item of suddenly, the new pattern can last for some least weasels, resulted in poor reproduction in decades, and the system can return to the old weasels, and the change in weasel abundance state again. and dynamics caused the change in rodent Hansson and Henttonen (1985a), dynamics. In other words, earlier cycles in and Hanski et al. (1991) related the gradient the bank vole were not caused by bank voles in Fennoscandian rodent cyclicity (increas­ themselves but were rather a side effect of ing amplitude and cycle length to north) to Microtus-Mustela interaction. In a way, bank the length ofwinter and snow depth. Ifsome­ voles were alternative prey for weasels. Han­ thing drastic had happened in the winter cir­ ski and Henttonen (1996) and Oksanen cumstances in the north, the resulting change and Henttonen (1996) have discussed the in animal community structure could well problem further. Models with a specialist have been reflected in the rodent dynamics, predator and two competing prey species, the too, but there is no real evidence ofsuch envi­ dominant ofwhich is the preferred prey, yield ronmental change. The role ofthe introduced patterns similar to those observed at Pallas­ American mink has been mentioned (0 k­ jarvi (Hanski and Henttonen 1996). sanen and Henttonen 1996), but does not The essence of this model is that in the seem strong enough, especially in the long present more stable situation, the predation snowy winter as at Pallasjarvi. Hansson by stoat in summer time is enough to prevent (1999) suggested that the change is due to the the increase of the preferred prey items, the landscape change, especially the change in larger and clumsier Microtus, C. rufocanus modern forestry. Briefly, the change in the and also lemmings in the birch forest, while age structure of boreal forests has resulted stoats are maintained in winter by smaller less food for forest rodents and also lower Clethrionomys (C. glareolus in taiga and C. densities in open habitats in the increase rutilus in the birch forest). Predation by stoat phase. Earlier the dispersal by forest rodents 92 H. Henttonen

into open habitats in the early increase phase Julliard et al. 1999), but there are always buffered predation on Microtus which gave some C. rutilus amongst the bank voles. In them potential to increase. This was followed some years also C. rufocanus has been quite by increase of least weasel and subsequent common. Because breeding Clethrionomys crash, as described above. In the present females have exclusive territories also be­ situation, predation on Microtus is not buff­ tween the species, the dynamics of competi­ ered, weasels do not increase, and dynamics tively dominant species (C. rufocanus) are re­ have changed to a different state. flected in the abundance of subordinate spe­ The core conclusion ofall ofthis is that cies. Notice, however, that these inverse the dynamics of any small mammal species competitive interactions take place in the should not be analysed out ofits community course ofincrease and peak years; during the context. deep crash phases all species are around zero It remains to be seen how long this "ex­ densities due to reasons related to cyclicity, ceptional" phase in Lapland will last. How­ not to interspecific competition. ever, the change in dynamics have several The length ofthe breeding season ofthe consequences. First of all, the high spring bank vole varies considerably at Pallasjarvi. densities characterising cyclic rodent peaks Males mature under snow in early April, 1.5 are now missing. The breeding of many months before snow melt, and first litters are predators and birds ofprey depends crucially born in early June. The variation in timing of on the high peak spring densities of rodents. the first litter is about two weeks. In high den­ It is probable that the numbers of some birds sity years young animals do not mature at all ofprey in Lapland have already declined. An­ but in the early increase phase (like 1972) other consequence of missing cyclic peak young females pregnant for the first time can densities and subsequent heavy grazing be found in still October. Young born later in pulses may be that the bottom and field layer summer usually delay their maturation to the on habitats favoured by voles and lemmings following spring. Based on the long-term live may change; species abundances may change trappings on the large grid, winter survival of if the most preferred plant species escape subadult females is clearly higher than sur­ heavy grazing for long periods (see also Vir­ vival of adult females (Prevot-Julliard et tan en et al. 1997). al. 1999). On the other hand, empirical values for the difference in survival were not consis­ tent with survival values suggested by theo­ 5. SOCIAL PATTERNS AND retical models if the delayed maturation was supposed to be optimal. Instead, our evidence INTERSPECIFIC COMPETITION suggests that delayed maturation is due to so­ cial constraints (P rev o t-J u 11 i ard et al. At Pallasjarvi, the bank vole densities 1999). There was a density-dependent rela­ can vary from zero to about 50-60 individu­ tionship between maturation rate of young als per hectare (Henttonen 1987, Hent­ voles and the density of established breeding tonen et al. 1987, Prevot-Julliard et al. females. Furthern1ore, Y occoz et al. (in 1999). During the cyclic crash phases the press) found marked seasonal structure in di­ snap trapping effort of 7000 trap night could rect density-dependence that was stronger in produce only 5 animals (spring 1979), and summer than in other seasons. during the deepest crashes (autumn 1979) the Communal nesting has often been sug­ snap trapping based on permanent small gested for bank voles in winter time, but our quadrats (over 2000 trap nights) could have unpublished analyses of the space use by been without any catch. On the other hand, bank voles in winter time do not support ag­ the highest densities ofClethrionomys (based gregated distribution, i.e. communal nesting. on live trappings on the large grid) after snow On the other hand, there is some variation in melt in spring have been around 10 breeding the space use among winters, and further­ voles/ha. Highest autumn densities have been more, the winter distributions are often more found in cyclic increase phases and some­ aggregated on the food grid. times during the stable period, and have been Our extensive field experiments (Jor­ around 50-60 Clethrionomyslha. The bank tikka 1990, Jortikka and Henttonen, un­ vole is clearly the dominating species in the publ.) showed that the abundance of C. ruti­ taiga forests at Pallasj arvi (Fig. 1, Hent­ lus is drastically limited by C. glareolus. tonen 1997,Henttonen eta!. 1987,Hent­ When the bank voles were removed, the den­ tonen and Hanski 2000, Prevot- sities ofred voles increased to the same level Long-term dynamics of the bank vole 93

300 • All Clethrionomys • C. gla 250 • C. rut '\/ Both species ..-en c 200 ·0- a. 0> 150 c ·a.- a. m 100 L.. t- 50

0 \}------'\/

June July June July June July 1977 1978 1981

Fig. 2. Early summer space use, expressed as the number of trap stations inhabited, by breeding Clethrionornys at various species abundances. In 1981 C. glareolus was alone on the grid (4.5 ha), in 1977 C. rufocanus had low density and both species could still expand their space use. In 1978 during high density of C. rufocanus, the area occupied by C. glareolus declined while that by C. rufocanus expanded. See also the same years in Fig. 1 as those ofbank voles in the controls or where Clethrionomys was about the same, in 1978 it red voles were removed. Removal of red was somewhat higher. In 1981 total Clethrio­ voles did not have any significant effect on nomys was C. glareolus alone. In 1977 both the bank vole abundance. Red voles are found species expanded their inhabited area from only in old taiga forest at Pallasjarvi while June to July, even though the number of trap bank voles are found on all habitats up to the station where both species were found, re­ tree limit of Norway spruce on the mountain mained minimal. However, in 1978 with slopes. The overlap zone of these two Cle­ highest density of C. rufocanus, the area in­ thrionomys species in western Lapland is habited by C. glareolus drastically declined 100- 150 km. Red voles have been found on when that of C. rufocanus expanded. Simul­ all permanent forest quadrats, but there seems taneously the overlap declined a bit. During to be a preference to drier habitats. Similarly, low to moderate densities of C. rufocanus, its on the large grid the few red vole territories impact on C. glareolus is not drastic, but with are usually found in the same, drier part ofthe the increasing density and space use of C. ru- grid. In winter time both the bank and red focanus, C. glareolus clearly loses ground. voles climb to trees to feed on lichens (Usnea This can also be seen in those years when the and Alectoria). The stomach analyses show grey-sided voles have increased towards the that e.g. in February the red voles rely almost end ofthe peak like in 1978 and 1983 (Fig. 1), exclusively on arboreal lichens whereas the when the density of bank voles has declined bank voles have usually also some green ma­ already in the course ofthe peak year. terial (Vace inium myrtillus shoots) in the The Pallastunturi mountain group is the their stomachs. southernmost area in western Lapland where The third Clethrionomys-species, the Lemmus lemmus is petrnanently found, and grey-sided vole is clearly larger and competi­ therefore Lemmus is a very interesting part of tively dominant over the two smaller conge­ the local rodent fauna. It should be remem­ nerics. Due to the teiTitorial behaviour in bered that the dynamics of Lemmus lemmus breeding Clethrionomys, the occasional differ drastically between south-central and abundance ofgrey-sided voles in taiga forests northern Fennoscandia. Instead ofthe regular reduces the space available for smaller spe­ 3-4-year cycles found in the southern Nor­ cies (Fig. 2). The number of trap stations on way, the pattern is much more irregular in the large grid inhabited by breeding individu­ Lapland (Henttonen and Kaikusalo als of either species in early June and July is 1993), where the lemming peaks coincide shown at different species abundances. In with those ofvoles, but there are not lemming 1977 and 1981 the total area inhabited by peaks during every vole peak. In Fig. 1, the 94 H. Henttonen others primarily refer to lemmings. Either were used, the difference was 1.5-2 times, lemmings totally miss some of the vole and when mouse chow was used, the differ­ peaks, or the increase in lemmings is so mi­ ence was 2-3 times higher (Henttonen et al. nor that we do not observe it. Furthermore, 1987, Prevot-Julliard et al. 1999). How­ lemming peaks occur more often in the north­ ever, on both treatments the seasonal density em mountain areas than further south in the variations were simultaneous. Increased den­ boreal zone. sities were mainly due to the higher number The great lemming outbreaks, with of breeding females per hectare on the food highly visible long-distance movements grids. Mouse chow also induced occasional south into the boreal taiga, show a long-terrn winter breeding in the bank vole which is not pattern (Henttonen and Kaikusalo 1993). otherwise seen at Pallasjarvi. These great lemming years, extending even I mentioned earlier the density­ south ofthe Arctic Circle in Finnish Lapland, dependent maturation of young. This have occurred 1970, 1938 (not quite to the AC), 1902-03, 1872, 1840, 1810, 1787, density-dependency was weaker on the large 1755. After the intensive movements into the food grid than on the control grid (Fig. 4 in taiga, lemmings seem to survive there for Prevot-Julliard et al. 1999). More young some time. After the 1970 large-scale move­ could probably obtain territories at a given ments, local lemming peaks were again ob­ density level of breeding females, obviously served in many areas in Lapland in rhythm because food addition allowed for smaller with the vole dynamics in 1974, 1978, and territories of breeding females. Even though 1982. For example in forests and mires at Pal­ food addition resulted in higher densities, its lasjarvi, lemmings were found in 1974 and effect on the density-dependent structure was 1978. During the outbreak years, lemmings negligible. Moreover, stability analyses of are the most abundant rodents in all habitats the deduced model suggested that the dynam­ in northern taiga, like in 1970 at Pallasjarvi ics were not affected by food addition. Varia­ (Fig. 1) (Henttonen et al. 1987, Hent­ tion in food resources is therefore probably tonen and Kaikusalo 1993), and have a not the cause for the multi-annual fluctua­ strong impcat on other syntopic vole species, tions in bank voles at Pallasjarvi (Y o cco z as well as on the vegetation (Kalela and Ko­ et al., in press). ponen 1971). Finally, based on my experience at Pal­ The periodic disappearance oflemmings lasjarvi, I would seriously warn about the po­ from the taiga has been explained by at least tential problems in live-trapping studies. It three ways (Henttonen and Kaikusalo 1993): has been a common practice in many studies 1) winter food supply in taiga is not satis­ to prebait live traps, or use considerable factory to lemmings, amounts of bait food in the traps or in trap 2) large and clumsy lemmings are more boxes. It is not known how much these sensitive to predation than sympatric voles, practices affect the food-related density­ 3) parasites (like Babesia) or pathogens dependent processes in vole populations. occurring in other rodent species in the taiga AsshownbyPrevot-Julliard eta/. (1999), could be harmful to lemmings. regulation of density-dependent maturation In the second and third case, bank voles in the bank vole is clearly affected by addi­ clearly have an advantage over lemmings through shared predators and/or parasites and tional food. In the first years of the food ex­ pathogens. Our unpublished results both from periment on the large food grid, we used no Pallasjarvi and Kilpisjarvi suggest that lem­ more than 300 kg oat seeds per year on 2.8 ha, mings are preferred by predators over smaller and the impact was clear. Rodents need only and agile voles of C. glareolus/rutilus type. 2-3 grams ofextra food per day to turn their energy balance positive, i.e. weight decrease turns to weight increase (my unpubl data). 6. EXPERIMENTS WITH Because of our intensive trapping schedule, SUPPLEMENTAL FOOD we use only 2-3 grams of bait in the trap on our control grids, and we do not throw the In the long-term experiment on the large bait trash onto the field but collect them in a control grid and large food grids, as well as in waste bag. It may be that most of the live­ replicated experiments, the Clethrionomys trapping studies in fact are unintentional ex­ densities were always higher on the food periments with supplemental food without grids than on the controls. When oat seeds controls. Long-term dynamics ofthe bank vole 95

7. P ARASITOLOGICAL RESEARCH demic theses - altogether some 150 items. The close connection of the protected na­ tional park and the experimental forests of Since the late 1970's intensive research Finnish Forest Research Institute in near vi­ on parasites of rodents and shrews has been cinity offers splendid opportunities for fur­ going on at Pallasjarvi with Voitto Haukisalmi. ther long-tetm research. At Pallasjarvi, the This work was started by faunistical and two main expectations are in mind: when do taxonomical analyses (e.g. Tenor a et al. the cycles return, and when is the next great 1983, 1985, Haukisalmi and Tenora lemming outbreak? 1993), but one ofthe main purposes has been to monitor the long-tertn dynamics of vole helminths in their fluctuating host popula­ tions, primarily in the bank voles (e.g. 9. REFERENCES Ha u k is a 1 m i et al. 1988, Ha u k is a 1m i and Hen t tone n 2000). The change in the Hansen T.F., Stenseth N .C., Henttonen H. 1999 vole dynamics has introduced an additional - Multiannual vole cycles and population regula­ interesting element into the parasite dynam­ tion during long winters: an analysis of seasonal density dependence - Am. Nat., 154: 129- 39. ics (Haukisalmi and Henttonen Hanski I. , Henttonen H. 1996 - Predation on com­ 1993a, Haukisalmi and Henttonen peting species: a simple explanation of complex 2000). The interacting role of moisture and patterns - J. Anim. Ecol., 65 : 220-232. earlier host density seems to be the critical Han ski I., Han sson L., Hen ttone n H. 1991 - Specialist predators, generalist predators, and the factors for helminth abundance (H auk - microtine rodent cycle - J. Anim. Ecol., 69: i sa 1 m i and Hen t tone n 1990). The role 353- 367. of rainfall is understandable because all the Hanski I., Henttonen H ., Hansson L. 1994 - Te­ helminths have stages in their life cycle that mporal variability of population density in micro­ have to survive outside ofthe host. We have tine rodents: a reply to Xia and Boonstra. - Am. Nat., 144: 329- 342. not been able to demonstrate that helminths Hanski I., Turchin P., Korpimaki E., Hentto­ affect the population dynamics of rodents nen H. 1993 - Population oscillations of boreal (Hau k is a 1m i and Hen t tone n 2000). rodents: regulation by mustelid predators leads to From the original work on long-term chaos - Nature, 364: 232- 235. Hansson, L. 1999 - Intraspecific variation in monitoring of the parasites dynamics in the dynamics: small rodents between food and bank vole and the possible impacts of para­ predation in changing landscapes - Oikos, 86: sites on rodents, the parasite research has ex­ 159- 169. panded to analyze various other aspects in the Hansson L., Henttonen H. 1985a - Gradients in population dynamics of helminths. These in­ cyclicity of small rodents: importance of latitude and snow cover - Oecologia (Berlin), 67: clude interspecific relation of helminth spe­ 394 402. cies in a very restricted space, the intestine of Ha n ss on L., Hen t tone n H. 1985b - Regional the host (Haukisalmi and Henttonen differences in cyclicity and reproduction in Clet­ 1993 b, c), and the general dynamic strategies hrionomys spp.: Are they related? - Ann. ZooI. Fenn., 22: 277- 288. ofso-called common and rare (core and satel­ Hansson L., Henttonen H. 1988 - Rodent dyna­ lite) helminth species of bank voles (Rent­ mics as community processes - Trends Ecol. tonen and Haukisalmi 1995, Hauki­ Evol., 3: 195- 200. salmi and Henttonen 1999). The last Hansson L., Henttonen H. 1989 - Rodents, preda­ mentioned aspect has a strong connection to tion and wildlife cycles - Finn. Game Res., 46: 26-33. biodiversity problems: a great proportion of Haukisalmi V., Henttonen H. 1990 - The impact existing species are parasites, and therefore of climatic factors and host density on the long-­ understanding of factors affecting the abun­ term population dynamics of vole helminths - dance and dynamics of parasites is of great Oecologia (Berlin), 83 : 309-315. Haukisalmi V., Henttonen H. 1993a - Popula­ importance. tion dynamics of Taenia polyacantha metacesto­ des in the bank vole Clethrionomys glareolus - Ann. Zool. Fenn., 30: 81- 84. 8. CONCLUSIONS Haukisalmi V., Henttonen H. 1993b- Co-exi­ stence in helminths of the bank vole Clethriono­ mys glareolus. I. Patterns of eo-occurrence - J. So far, the material ofPallasjarvi rodent Anim. Ecol., 62: 221- 229. Haukisalmi V., Henttonen H. 1993c - Co-existen­ project has been used in more than 50 refe­ ce in helminths of the bank vole Clethrionomys reed publications as well as in a number of glareolus. II. Intestinal distributions and interspe­ Finnish articles, congress abstracts and aca- cific interactions - J. Anim. Ecol., 62: 230-238. 96 H. Henttonen

Ha u k is a l m i V., Henttonen H. 1999 - Determi­ Henttonen H.,Wallgren H. 2000 - Small rodent nants of helminth aggregation in host populations: dynamics and communities in the birch forest individual differences or spatial heterogeneity? - zone of northern Fennoscandia (In: Mountain Ecography, 22:629-636. birch forest ecosystem, Eds. Wielgolaski F., So­ Haukisalmi V., Henttonen H. 2000. Variability of nesson M., Karlsson S., Laine K.) - Man and helminth assemblages and populations in the bank Biosphere Series, UNESCO. vole Clethrionomys glareolus - Polish J. Ecol., 48 J o rt i k k a A. 1990 - Metsamyyran Clethrionomys (Suppl.): 219-230. glareolus ja punamyyran C. rutilus lajien Haukisalmi V., Tenora. F. 1993- Catenoteania valisesta kilpailusta. M.Sc. Thesis, Dept. Zool., henttoneni n. sp. (Cestoda, Catenotaeniidae), a pa­ Univ. Helsinki. rasite of voles Clethrionomys glareolus and C. ru­ Kalela 0., Koponen T. 1971 - Food consumption tilus in Finnish Lapland. - Folia Parasitol., and movements of the Norwegian lemming in are­ 40:29- 33. as characterized by isolate fells - Ann. Zool. Haukisalmi V, Henttonen H., Tenora F. 1987- Fenn., 8: 80- 84. Parasitism by helminths in the grey-sided vole Korpimaki E., Norrdahl K. 1998 - Experimental Clethrionomys rufocanus in northern Finland: reduction of predators reverses the crash phase of influence of density, habitat and sex of the host - small rodent cycles - Ecology, 79: 2448- 2455. J. Wildl. Dis., 23: 233- 241. Myllymaki A., Paasikallio A., Pankakoski E., Haukisalmi V., Henttonen H., Tenora F. 1988 - Kanervo V. 1971 - Removal experiments on Population dynamics of common and rare hel­ small quadrats as a means of rapid assessment of minths in cyclic vole populations - J. Anim. Ecol., the abundance of small mammals - Ann. Zool. 57:807- 826. Fenn., 8: 177- 185. Haukisalmi V., Henttonen H., Vikman P. 1996 - Norrdahl K. , Korpimaki E. 1995 - Mortality fac­ Variability of sex ratio, maiting probability and tors in a cyclic vole population. - Proc. Royal egg production in an intestinal nematode in its flu­ Soc. London (B), 261: 49- 53. ctuating host population - Int. J. Parasitol., 26: Oksanen T., Schneider M., Rammul U., 755-764. Hamback P.A., Aunapuu M. 1999 - Popula­ Henttonen H. 1985 - Predation causing extended tion fluctuations of voles in North Fennoscandian low densities in microtine cycles: further evidence tundra: contrasting dynamics in adjacent areas from shrew dynamics - Oikos, 45: 156-157. with different habitat composition - Oikos. 86: Hen tt onen H. 1987 - The impact of spacing beha­ 463- 78 vior in microtine rodents on the dynamics of le­ Oksanen T., Henttonen H. 1996 - Dynamics of ast weasels Mustela nivalis - Oikos, 50: voles and small mustelids in the taiga landscape of 366- 370. northern Fennoscandia in relation to habitat quali­ Henttonen H. 1997 - The rodent project at ty - Ecography, 19: 432-443. Pallasjarvi (In: Pallas-Symposium 1996, Eds. Prevot-Julliard A.-C., Henttonen H.. Yocco z Loven L., Salmela S.) Metsantutkimuslaitoksen N.G ., Stenseth N.C. 1999 - Delayed maturation tiedonantoja, 623 : 49- 58. in female bank voles: optimal decision or social Henttonen H., Hansson L. 1984- Interspecific re­ constraint? - J. Anim. Ecol., 68: 684--697. lations among small rodents in European boreal Stenseth N.C. 1999 - Population cycles in voles and and subarctic environments - Acta Zool. Fenn., lemmings: density dependence and phase depen­ 172: 61--65. dence in a stochastic world Oikos, 87: 427 61. Henttonen H., Hanski I. 2000 Population dynamics Tenora F., Henttonen H., 1-laukisalmi II. 1983 - of small rodents in northern Fennoscandia (In: On helminths of rodents in Finland - Ann. Zool. Chaos in Real Data. The Analysis of Non-Linear Fenn., 20: 37-45. Dynamics from Short Ecological Time Series, Tenora F. , Haukisalmi V., Henttonen H. 1985 - Eds. Perry J.N., Smith R.H., Woiwod I.P., Morse Andrya kalelai sp.n. and (?)Anoplocephaloides D.) - Kluwer Academic Publishers, Dordrecht. sp., Cestoda, Anoplocephalidae, parasites of Cleth­ pp. 73-96. rionomys-rodents in Finland - Ann. Zool. Fenn., Henttonen H., Haukisalmi V. 1995 - Loiset 22: 411-416. luonnon monimuotoisuuden indikaattoreina Vi ita I a J. 1977 - Social organization in cyclic (In: Monimuotoisuus metsien hoidossa. subarctic populations of the voles Clethrionomys Metsantutkimuslaitoksen tiedonantoja, Eds. Han­ rufocanus (Sund.) and Microtus agrestis (L.) - nelius S., Niemela P.) 564: 73- 84. Ann. Zool. Fenn., 14: 53- 93. Henttonen H., Kaikusalo A. 1993 -Lemming Virtanen R., Henttonen H., Laine K.M. 1997 - movements. (In: The biology of lemmings, Eds. Lemming grazing and the structure of snowbed Stenseth N.C., Ims R.A. ). Academic Press for the plant community - a long-term experiment at Linnean Society, 157- 186. Kilpisjarvi, in Finnish Lapland Oikos, Henttonen H., Kaikusalo A., Tast, J., Viitala 79: 155- 166. J. 1977 - Interspecific competition between small Yoccoz N.G., Stenseth N.C., Henttonen H. , rodents in subarctic and boreal ecosystems - Prevot-Julliard A.-C. (In press) - Food addi­ Oikos, 29:581- 590. tion does not affect population dynamics in small Henttonen H., Oksanen T., Jortikka A., Hau­ rodents: effects on the seasonal density-dependent k is a I m i V. 1987 - How much do weasels shape structure on the bank vole ( Clethrionon1ys glare o­ microtine cycles in the northern Fennoscandian fus) in a long-term experiment in northern Fin­ taiga? - Oikos, 50:353- 365. land. - J. Anim. Ecol.

(Received after revising June 2000)